image repair on existing products

app/services/produce_reference.py

@@ -126,9 +126,46 @@ def usda_fdc_id(product_name: str) -> Optional[int]:
     return entry.get("usda_fdc_id") if entry else None
 
 
+# Words that make a name a DIFFERENT PRODUCT from the commodity it starts with,
+# rather than a variety of it.
+#
+# `lookup` falls back through shorter leading prefixes, which is what lets
+# "Mango Totapuri" and "Banana Robusta" find their base commodity - varieties of
+# the same thing, correctly sharing one photo. The same fallback turns "Coconut
+# Oil" into "Coconut" and "Apple Cider Vinegar" into "Apple", and a bottle of
+# oil is not a variety of coconut. Handing it a photo of the raw fruit is the
+# same class of error as the apple-juice image this table was built to fix, just
+# pointing the other way.
+#
+# No row in the catalogue trips this today - all 159 names are exact keys - so
+# this guards the case a future upload introduces, which is precisely when
+# nobody would be looking.
+_DERIVED_PRODUCT_WORDS = frozenset({
+    "oil", "vinegar", "juice", "squash", "syrup", "jam", "jelly", "sauce",
+    "ketchup", "puree", "paste", "pickle", "powder", "flour", "atta", "rava",
+    "flakes", "chips", "crisps", "candy", "extract", "essence", "milkshake",
+    "smoothie", "dried", "fried", "roasted", "pickled", "canned", "frozen",
+})
+
+
 def image_url(product_name: str) -> Optional[str]:
+    """The reviewed photo for a commodity, or None.
+
+    Stricter than `lookup` on purpose, and only here: `usda_fdc_id` keeps the
+    plain prefix fallback, because this guard is about not showing a misleading
+    PICTURE. Nutrition for a derived product is refused further upstream by the
+    category dispatch in `nutrition_data_service`.
+    """
     entry = lookup(product_name)
-    return entry.get("image_url") if entry else None
+    if not entry:
+        return None
+
+    matched = _normalize(entry.get("product_name") or "")
+    leftover = set(_normalize(product_name).split()) - set(matched.split())
+    if leftover & _DERIVED_PRODUCT_WORDS:
+        return None
+
+    return entry.get("image_url")
 
 
 def all_entries() -> Dict[str, Dict[str, Any]]: